Completed exercises (date returned)	Additional exercises - optional, will discussed @ extra study session
6.1 - 6.4, 6.17 - 6.19 (3/2)	6.6, 6.7, 6.8, 6.20
7.1, 7.2, 7.5 (3/9)	none
8.5 - 8.8 (5/4)	8.1 - 8.4, 8.9
9.10 - 9.19 (3/9)	Studying-A-Study Practice
10 (quiz 4/6)	10.1, 10.2
11.3, 11.4 (5/4)	11.2, 11.7
12 (quiz 5/4)	to be determined

Case Counts
- Prevalence count = no. of cases at particular time
- Incidence count = no. of case onsets that accumulate over period of study
- Counts not useful without context � convert counts to rates or proportions
Incidence Proportion (synonyms: risk, cumulative incidence)
- Numerator = onsets
- Denominator = no. @ risk
- Closed populations only
Incidence Rate (synonyms: incidence density, person-time rate)
- Numerator = onsets
- Denominator = person-time @ risk
  - In a cohort: sum person-time individually
  - In an open populations: Person-time approx. = (average pop. size) � (time of observation)
    Reason given on p. 131
  - Example of a rate in a cohort [line drawings showing "survival" curve is integrating concept]
- Interpretation: a) speed, b) inverse of wait time b) risk
Prevalence
- Numerator = prevalence count
- Denominator = population size
- Dynamics of Prevalence � Figure 6.9 on p. 185
  - Inflow = incidence
  - Outflow = resolution rate (death and recover), which is related to average duration
Use of multiplier to report rates does not change value (but does makes it easier to interpret)
See Chapter for details

Goal: to reduce distortions, to encourage �like-to-like� comparisons (sissection of numerical illustrations could reveal much
Confounding by age (as an illustration)
- Illustrated by Florida / Alaska example (pp. 146 � 147)
- Florida has higher crude mortality rate b/c age (extraneous factor) confounds relation
Terminology
- �Rate� � any incidence or prevalence (economy of language)
- Study population - the population for which rates will be adjusted
- Reference population - the population which serves as an age standard
- Crude rate � rate for entire population
- Strata-specific rate - rate within subgroup
- Adjusted rate � rate for entire group mathematically compensated for extraneous factor such as age
Two methods of adjustment
- Direct
- Indirect (via the SMR) -- no calculations
Formulas and notation
- Subscript indicates strata number
- Cap letters indicate value comes from reference population
- Small letters indicates value comes from the study population
Adjustment for multiple factors
- e.g., 5 age groups, 2 genders, 3 ethnic groups result in 5 � 2 � 3 age-gender-ethnic strata
- Method applied to strata
See Chapter for details

Premise
- Divide population into two groups
  - Exposed (group 1)
  - Non-exposed (group 0)
- If risk is higher in one group or the other, we say there in an association between E and D
Comparing Risks
- Suppose R₁ = 2% and R₀ = 1%
- Absolute comparison: Risk Difference = 2% � 1% = 1%
- Relative comparison: Relative Risk = 2% � 1% = 2
RR vs. RD
- RR measures strength of association (good for making causal statements)
- RD measures excess risk in absolute terms (good for making public health impact statements)
RR & RD are related
- Algebraic expression
- Concept: a large multiplier (RR) of a small baseline risk (R₀) produces few additional
  cases (small RD)
Multiple levels of exposure
- Compare each exposure level to the baseline (�non-exposed�) rate
- Allow for determination of dose-response relations
Now see Chapter

General concepts
- Random error = chance error = lack of precision (imprecision)
- Systematic error = bias = lack of validity
Random error
- Derives from chance difference and random sampling
- Decreases as sample size increases
- Addressed with confidence intervals and p values
Systematic error - a matter of design
- Selection bias
- Information bias
- Confounding
See chapter